§85

A signpost doesn't signpost how it is to be taken. So you might say that they leave room for doubt - the can be misinterpreted.

"So I can say that the signpost does after all leave room for doubt. Or rather it sometimes leaves room for doubt and sometimes not. And this is no longer a philosophical proposition, but an empirical one."

"The signpost leaves room for doubt" is a 'philosophical proposition' then. I assume this means that it is a grammatical proposition - it points to an internal/conceptual relation. - If something counts as a signpost then it is the kind of thing taht might be misinterpreted (we might have doubts about how to follow it).

BUT "it [the signpost] sometimes leaves room for doubt and sometimes not" is empirical. - As a matter of fact people sometimes have doubts about how a sign is to be taken and sometimes they are quite certain that it is to be taken in a particular way.

(Here, when Wittgenstein is talking about signposts, he also, I imagine, has rules in mind - that a rule is like a signpost).